HAL Id: hal-02506764
https://hal.umontpellier.fr/hal-02506764
Submitted on 12 Mar 2020
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of
sci-entific research documents, whether they are
pub-lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Distributed under a Creative Commons Attribution - NonCommercial| 4.0 International
Regulatory Divergence between Parental Alleles
Determines Gene Expression Patterns in Hybrids
Marie-Christine Combes, Yann Hueber, Alexis Dereeper, Stéphanie Rialle,
Juan-Carlos Herrera, Philippe Lashermes
To cite this version:
Marie-Christine Combes, Yann Hueber, Alexis Dereeper, Stéphanie Rialle, Juan-Carlos Herrera, et
al.. Regulatory Divergence between Parental Alleles Determines Gene Expression Patterns in Hybrids.
Genome Biology and Evolution, Society for Molecular Biology and Evolution, 2015, 7 (4),
pp.1110-1121. �10.1093/gbe/evv057�. �hal-02506764�
Regulatory Divergence between Parental Alleles Determines
Gene Expression Patterns in Hybrids
Marie-Christine Combes
1,*, Yann Hueber
1, Alexis Dereeper
2, Ste´phanie Rialle
3, Juan-Carlos Herrera
4, and
Philippe Lashermes
11IRD, UMR DIADE, Montpellier Ce´dex 5, France 2IRD, UMR IPME, Montpellier Ce´dex 5, France 3
MGX-Montpellier GenomiX, Institut de Ge´nomique Fonctionnelle, Montpellier Ce´dex 5, France
4
Centro Nacional de Investigaciones de Cafe, CENICAFE – FNC, Manizales, Colombia *Corresponding author: E-mail: marie-christine.combes@ird.fr.
Accepted: March 18, 2015
Data deposition: This project has been deposited at European Nucleotide Archive under the accession PRJEB7565.
Abstract
Both hybridization and allopolyploidization generate novel phenotypes by conciliating divergent genomes and regulatory networks in the same cellular context. To understand the rewiring of gene expression in hybrids, the total expression of 21,025 genes and the allele-specific expression of over 11,000 genes were quantified in interspecific hybrids and their parental species, Coffea canephora and Coffea eugenioides using RNA-seq technology. Between parental species, cis- and trans-regulatory divergences affected around 32% and 35% of analyzed genes, respectively, with nearly 17% of them showing both. The relative importance of trans-regulatory divergences between both species could be related to their low genetic divergence and perennial habit. In hybrids, among divergently expressed genes between parental species and hybrids, 77% was expressed like one parent (expression level dominance), including 65% like C. eugenioides. Gene expression was shown to result from the expression of both alleles affected by intertwined parental trans-regulatory factors. A strong impact of C. eugenioides trans-regulatory factors on the upregulation of C. canephora alleles was revealed. The gene expression patterns appeared determined by complex combinations of cis- and trans-regulatory divergences. In particular, the observed biased expression level dominance seemed to be derived from the asymmetric effects of trans-regulatory parental factors on regulation of alleles. More generally, this study illustrates the effects of divergent trans-regulatory parental factors on the gene expression pattern in hybrids. The characteristics of the transcriptional response to hybridization appear to be determined by the compatibility of gene regulatory networks and therefore depend on genetic divergences between the parental species and their evolutionary history.
Key words: hybridization, cis- and trans-regulation, allele-specific expression, allopolyploidy.
Introduction
Hybridization and allopolyploidization, both prominent events in plant evolution, generate novel hybrid phenotypes (Hegarty and Hiscock 2009). Changes in gene expression have been observed in hybrids that reconcile divergent genomes and regulatory networks in the same cellular context (Landry et al. 2007). Elucidating the regulatory changes at the origin of the variability of gene expression in hybrids is a key to un-derstanding gene-expression novelty and its role in adaptive evolution (Wolf et al. 2010;Buggs et al. 2014). In recent years, several genome-wide studies have investigated gene regulatory divergences and/or inheritance of gene expression
in hybrids and allopolyploids (Chague´ et al. 2010;McManus et al. 2010;He et al. 2012;Qi et al. 2012;Shi et al. 2012;Bell et al. 2013;Yoo et al. 2013;Cox et al. 2014). In these studies, cis and trans effects have been identified in gene expression differences between parental alleles or homoeologs (McManus et al. 2010;He et al. 2012;Shi et al. 2012;Bell et al. 2013; Xu et al. 2014). In addition, nonadditive gene expression variations in hybrids or allopolyploids relative to parental expression level have been shown. In particular, dom-inance expression was observed for numerous genes, the ex-pression level in the hybrid or in the allopolyploid being similar to that exhibited by one of the parent, up or down regulated
GBE
ßThe Author(s) 2015. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com
relative to the other parent. Furthermore, for most of these studies, the genome-wide expression level dominance was skewed toward one parent (termed as biased expression level dominance;Grover et al. 2012). This phenomenon ap-pears to be common and was reported in various organisms (flies, yeast, plants) (McManus et al. 2010;Qi et al. 2012;Bell et al. 2013;Yoo et al. 2013;Cox et al. 2014). Although overall patterns in gene expression changes in hybrids were described, the relationships between inheritance of gene expression and gene regulatory divergences have not been completely ex-plored and the genetic bases of the variations in allelic expres-sion are poorly understood. To understand the rewiring of gene expression in hybrids, we investigated interspecific hy-brids between two Coffea species, Coffea canephora (C) and Coffea eugenioides (E). The two species are perennial woody trees and display considerable variation in morphology, size, and ecological adaptation. Nevertheless, they are closely re-lated, display low sequence divergence (i.e., 1.3% average difference for genes) with very high gene synteny (Cenci et al. 2012), and are characterized by similar genome sizes, 710 Mbp for C and 660 Mbp for E (Noirot et al. 2003). In addition, they hybridize readily (Louarn 1976), and are parental species of the cultivated natural allotetraploid Coffea arabica (Lashermes et al. 1999). To understand the relationships be-tween the inheritance of gene expression and allele-specific expression (ASE), using RNA-seq technology, we analyzed in-heritance of gene expression, cis- and trans-regulatory diver-gences, and variations in ASE in the hybrids.
Materials and Methods
Plant Materials
Interspecific hybrids were generated between C, widespread species in West and Central Africa (0–1,500 m asl.), and E, endemic species in East Africa at much higher altitudes (1,000–2,200 m asl.). The hybrids were produced by crossing C, used as the female parent, with E (C E) and reciprocally using E as the female parent (E C) following the standard hybridization technique (Couturon et al. 1998). All the plants belong to the core collection of the Colombian National Center of Coffee Research (CENICAFE:www.cenicafe.org). For this study, we selected 17 plants including three plants of C. canephora species, four plants of C. eugenioides species, six plants of the F1 interspecific hybrids C E, and four plants of the reciprocal F1 crosses E C. All plant materials were grown in common garden growing conditions at Naranjal experimen-tal station (Chinchina, Colombia). Special care was taken to sample, young and just fully expanded leaves on each plant and to frozen the material immediately in liquid nitrogen. RNA Extraction and RNA Sequencing
Total RNA was extracted using the Qiagen RNeasy kit (Qiagen, Stanford, CA) according to the manufacturer’s instructions.
The quality and concentration of extracted RNAs were deter-mined using the Agilent DNA 1000 (Agilent, Santa Clara, CA). The messenger RNA (mRNA) libraries were constructed with the Illumina “RNA-seq sample prep” kit (Illumina, San Diego, CA) and sequenced separately on a Hiseq 2000 (Illumina) at the MGX platform (Montpellier Genomix, France). After qual-ity filtering (Phred score higher than 28 and removal of reads less than 50 bp), a total of 375 million single-end reads (~72 nt) were retained (ranging from 12 to 31 million reads per library), including 80 million, 73 million, 135 million, and 87 million for the C. canephora, C. eugenioides, C E hybrid, and E C hybrid libraries, respectively.
RNA-seq Data Processing
The data concerning the parents and the reciprocal interspe-cific hybrids were treated using the same method. Owing to the low genetic divergence between parental species, the 72-nt reads of each library were mapped to a C. canephora-coding DNA sequence reference (25574 CDS) as transcrip-tome reference (Denoeud et al. 2014) using Burrows Wheeler Aligner MEM (BWA MEM) (Li and Durbin 2010) with the default parameters. The RSeqC v2.4 software pack-age was used to check the quality of RNA-seq data from BAM files (.bam) (Wang et al. 2012). Gene coverage profiles that were slightly biased toward the 30
extremity of the genes were observed uniformly in all the samples used in the study. As previously (Combes et al. 2013) descrided, the gene expres-sion estimates for the E parent were assumed to be slightly underestimated, due to a slight technical bias caused by the nature of the transcriptome reference. The aligned sequences of each library were then analyzed to find SNPs (single nucle-otide polymorphism) with the GATK toolkit (http://www.
broadinstitute.org/gatk/) using the Unified-Genotyper
module with default parameters to obtain a list of SNPs and allelic data, and the depth-of-coverage module to obtain in-formation on depth coverage. To avoid artifacts due to reads from pseudogenes or repeat sequences, only the CDS identi-fied as single copy was used for subsequent analyses. A spe-cies-specific database of 82.916 SNPs, representing 14.206 CDS, was produced by retaining only divergent polymorphic nucleotide sites at which accessions of both parents were ho-mozygous for differences.
Differential Expression Analysis
Mapped reads of each sample were analyzed with the “DESeq” package (Anders and Huber 2010) in R software version 3.0 (R Foundation for Statistical Computing, Vienna, Austria).To remove the negative effect of background expres-sion noise on differential expresexpres-sion estimation, we restricted analysis to genes with minimum cumulated read counts (de-pending on the number of replicates) of 30, 40, and 60 for the C. canephora samples, the C. eugenioides, and E C and C E hybrid samples, respectively, in at least one parental
Regulatory Divergence between Parental Alleles
GBE
and hybrid group. For genes whose allele was not detected in interspecific hybrids, we changed the number of mapped reads from 0 to 1 in order to calculate ratios and perform statistical analyses (Marioni et al. 2008). Data were first nor-malized on the total number of counts with the DESeq pack-age (using “per condition” and “maximum” as arguments), taking the variance and the mean of the biological replicates into account. Next, pairwise comparisons were performed of the total level of gene expression of the parental species, and those of the hybrids with each of the parental species (nega-tive binomial test). The distribution of P values was checked for a false discovery rate using the Benjamini and Hochberg method at a = 0.05.
Inheritance Classifications
Expression inheritance was determined for differentially expressed genes in hybrids. Log-transformed and DESeq-normalized expression values of each parent were subtracted from those of the hybrid to examine changes in expression. As described byMcManus et al. (2010),genes whose total ex-pression in hybrids deviated more than 1.25-fold from that of either parent were considered to have nonconserved inheri-tance. Based on the magnitude and the direction of the changes, the genes were classified as displaying additivity (with E expression lower or higher than C expression), E or C expression level dominance (lower or higher than either parent: down and up), and transgressivity (lower or higher than both parent: down and up).
ASE and Assignment of cis- and trans-Regulatory Divergence
Using the species-specific SNP database, the reads of hybrids with SNPs were sorted into C or E allele-specific bins using Custom Perl Scripts. Genes with more than 5% of reads ex-hibiting ambiguous assignment (not consistent with the spe-cies-specific SNP database) due to the potential presence of paralogs were discarded (around 1.5% of genes). To examine the possibility of biased ASE, genes with low coverage (cumu-lative numbers of C-specific reads and E-specific reads was 80) were filtered out. ASE of replicates of each hybrid and parental expression of replicates of parents were then treated and normalized using the DESeq package. The count data were normalized on the total number of counts taking the variance and the mean of the biological replicates was taken into account. Relative allelic expression (RAE) corresponding to allele-specific read counts (Chyb or Ehyb)
among the total read counts (Chyb+ Ehyb) was determined
and expressed as the percentage of the Chyballele (%Chyb)
in the total gene expression of the hybrid.
At the transcription level, gene expression is governed by interactions between cis- and trans-regulatory factors. In hy-brids, both parental alleles are in the same cellular context and exposed to a common set of trans-regulatory factors; allelic
expression can therefore be altered due to cis- and/or trans-regulatory divergence between parental species. ASE in the hybrids compared with allelic expression in their parents makes it possible to distinguish between cis- and trans-regu-latory divergences. Several tests were performed (Wittkopp et al. 2004; Shi et al. 2012), first to compare the ratio of expression of the two parental alleles in the hybrids with the relative expression of the same alleles in their parents (two-sided prop.test in R, H0: C/E = Chyb/Ehyb, P value adjusted for
multiple testing; Benjamini–Hochberg method) and then to compare the RAE with the balance parental expression (two-sided prop.test in R, H0: Chyb/Ehyb= 1, P value adjusted
for multiple testing; Benjamini–Hochberg method). Based on these tests, genes were categorized as conserved, or as be-longing to cis-regulatory, regulatory, or cis- + trans-regulatory divergence categories. A similar and balanced allelic expression between parents and hybrids indicated conserved regulation, whereas a conserved unbalanced allelic expression between parents and hybrids was the signature of parental cis-regulatory divergences, and a balanced allelic expression only in hybrids revealed parental trans-regulatory divergences. For the remaining genes, parental expression divergences were the result of a combination of cis and trans effects. Compensating and enhancing cis and trans interactions were identified by comparing the directions of cis and trans effects. A compensating cis + trans interaction was inferred if the cis and trans effects of one gene were in the opposite direction (Chyb/Ehyb>1 and C/E < Chyb/Ehyb or Chyb/Ehyb<1
and C/E > Chyb/Ehyb). If the cis and trans effects of one gene
acted in the same direction (Chyb/Ehyb>1 and C/E > Chyb/Ehyb
or Chyb/Ehyb<1 and C/E < Chyb/Ehyb), then the interaction was
categorized as enhancing cis + trans interaction. Up- and Downregulation of Hybrid Alleles
For genes whose ASE was available, the expression regulation of both alleles was evaluated. For each allele, the variation in expression in the parents and in the hybrid was plotted on a graph (Chyb/C on the x axis and Ehyb/E on the y axis). Because
for each allele, the genome-specific cis-regulator factors are similar between the parental and hybrid cellular contexts, this graph provides information about changes in allelic expression due to trans-regulatory factors of the other genome. The po-sition of each gene depends on the up- and downregulation of both alleles and on the origin of trans-regulatory effects. Although negative values of the log2 transformed
allele-specific ratio (Chyb/C or Ehyb/E) indicate downregulation of
alleles in hybrids, positive values indicate upregulation. Analysis of Functional Enrichment of Gene
Ontology (GO)
Computational mapping and plant GO-slim annotation were performed using BLAST2GO PRO software (Conesa et al. 2005;Conesa and Gotz 2008) with default parameters for
Combes et al.
GBE
25.574 CDS of the C. canephora reference. For the genes selected at different steps of the analysis, molecular and cel-lular functions were searched for using an enrichment test (False Discovery Rate (FDR) < 0.05) to evaluate statistical significance.
Results and Discussion
Differential Gene Expression among Parental Species Before studying gene expression patterns in hybrids, we first characterized pre-existing differential gene expression be-tween the two parental species; 33% of the 21,052 genes we examined were significantly differentially expressed between parental species. Although in E, the gene expres-sion was slightly underestimated due to the use of the C transcriptome as mapping reference (Combes et al. 2013), 59% of differentially expressed genes were more highly ex-pressed in E than in C (fig. 1). In addition, after setting a threshold (log2 fold change ±1.25) so as to consider only
markedly differentially expressed genes, the percentage of differentially expressed genes between parents was reduced to 23%, but 60% of differentially expressed genes was still more highly expressed in E than in C. Thus, the estimate of the level of expression divergence between low genetic di-vergent Coffea species is slightly higher than those between intraspecific Cirsium parents (Bell et al. 2013) and consider-ably lower than those between divergent Drosophila
interspecific parents (McManus et al. 2010). In the GO term enrichments of differentially expressed genes, the pro-cesses of “regulation” and “response to stimulus” were over represented, whereas the “metabolic process” and “bind-ing” function were under represented (supplementary table S1,Supplementary Materialonline). GO term enrichments of the most expressed genes in each parental species were also identified, whereas enrichments of GO terms linked with binding function, metabolic process, “regulation of meta-bolic process” and “development” process (supplementary table S2,Supplementary Materialonline) were observed in C. GO terms associated with response to stimulus appeared over represented in E (supplementary table S3,
Supplementary Material online). These data suggest that
the expression profiles of the parental species are quite sim-ilar for basic cellular functions, but differ for genes involved in the regulation of metabolic processes and in the response to stimulus. However, more differences between both species may exist, indeed subtle changes (less than log2 fold
change ± 1.25) for some parental genes that could consider-ably affect the phenotype were not considered in the anal-ysis. Nevertheless, the observed transcriptomic divergences for genes involved in the regulation of metabolic processes and in response to stimulus are not surprising for Coffea species which result from a recent and rapid speciation (Cros et al. 1998; Anthony et al. 2010). Under different environmental pressure conditions, each species could have rapidly evolved by regulating gene expression leading to morphological variations and ecological adaptations as observed for species resulting from rapid adaptive radiation (Barrier et al. 2001;Kapralov et al. 2013). The role of gene expression variation in rapid evolution has already been sug-gested for Senecio aethnensis and Senecio chrysanthemifo-lius, two closely related species that recently diverged and are adapted to different altitude environments. In these spe-cies, regulatory differences in genes involved in environmen-tal adaptation appear to determine their phenotypic differences (Chapman et al. 2013;Muir et al. 2013). In ad-dition, variations in gene expression inducing changes in cel-lular function with relatively little change in genomic sequence are part of evolutionary processes (Whitehead and Crawford 2006;Ames and Lovell 2011).
Global Pattern of Gene Expression in Hybrids Biased Toward Coffea eugenioides
We characterized changes in gene expression in ten hybrids resulting from distinct hybridization events between acces-sions of C and E (3 and 4, respectively), and subdivided them into two groups according to the crossing direction (C E and E C). To assess the reproducibility of the results, we examined the two groups separately throughout the study. In all comparisons of the total level of gene expression of the hybrids with those of each parental species, we
FIG. 1.—Differences in gene expression between Coffea canephora
and Coffea eugenioides. The histogram shows the direction and magni-tude of changes in expression in genes exhibiting divergent parental ex-pression. Negative values indicate higher expression in C. eugenioides (E > C), and positive values indicate higher expression in C. canephora (C > E). A total of 33% of genes were differentially expressed between parental species (6,966 genes among the 21,052 genes analyzed), regard-less of the magnitude of the change (DESeq analysis, seesupplementary Methods,Supplementary Materialonline). Of these genes, significantly more (59%, n = 4,136) were highly expressed in C. eugenioides than in C. canephora (41%, n = 2,830) (E/C > 1), (binomial test, P < 0.001).
Regulatory Divergence between Parental Alleles
GBE
observed fewer differentially expressed genes relative to the E parent than relative to the C parent. Moreover, for each pair-wise comparison between hybrids and one of the parents, most differentially expressed genes were more highly ex-pressed in the hybrids than in the parental species, providing evidence for changes in the regulation of gene expression (mainly upregulation) after the genome merger that resulted from hybridization (fig. 2).
Inheritance of Gene Expression in Hybrids and Expression Level Dominance of Coffea eugenioides
Changes in gene expression in hybrids were also evidenced by classifying gene expression inheritance patterns. Indeed, in both groups of hybrids, among divergently expressed genes between parental species and hybrids (around 27%), we ob-served all the different forms of altered gene expression (ad-ditivity, up and down dominance, and up and down transgressivity) (fig. 3). Although around 13% and 10% of genes were binned in additivity and transgressivity categories, respectively, the majority of categories contained expression level dominance (the remaining 77%) and the E expression level dominance “up” accounted for almost half the genes. Although we applied different methods of gene classification, the distribution of genes in categories was close to that pre-viously reported for hybrids and allopolyploids (McManus et al. 2010; Qi et al. 2012; Bell et al. 2013; Yoo et al. 2013). Furthermore, the expression level dominance where the total expression of homeologs for a given gene in an allopoly-ploid is statistically equivalent to the expression level of that
gene in only one of the parents has been described many times before (Grover et al. 2012), but the molecular mecha-nisms of this phenomenon are still poorly understood.
Contribution of cis and trans Effects to Gene Expression Divergence
For genes with species-specific SNPs, RAE was determined and expressed as the percentage of the Chyballele (%Chyb)
in the total gene expression of the hybrid (supplementary fig. S1,Supplementary Materialonline). In all the genes we analyzed, both alleles were always expressed in both groups of hybrids and 70% contributed equally to the transcrip-tome of hybrids, that is, the average percentage of the Chyb
allele was around 50%. For genes showing unbalanced RAE, the Chyb allele was no more frequently expressed
than the Ehyballele. For C. arabica, the genome-wide
anal-ysis of the relative homoeologous gene expression has also revealed that neither of the two subgenomes was prefer-entially expressed (Combes et al. 2013).
In hybrids, gene expression differences between parental alleles can result from cis- and/or trans-regulatory changes. To elucidate the genetic basis of gene expression divergence, we compared the ASE in parental and hybrid samples to identify cis- and trans-regulatory divergences (fig. 4A). The RAE in hy-brids, in which parental alleles and regulatory networks inter-act in the same cellular environment, provides readout of relative cis-regulatory divergences. Allelic expression divergences not explained by cis-regulatory divergences be-tween the parental species are inferred to be caused by
FIG. 2.—Differentially expressed genes in each possible comparison between parents and hybrids (DESeq analysis, see supplementary Methods,
Supplementary Materialonline). Numbers in bold indicate the total number and fraction of genes differentially expressed in each comparison among the 21,052 genes analyzed. For each comparison, the number and fraction of upregulated genes in each group of plants are indicated at the ends of the arrows. In pairwise comparisons between gene expression in the hybrids and their parental species, the percentages of genes showing differential expression between hybrids and Coffea eugenioides (4.1% and 5.4 % for the comparison E C versus E and C E versus E, respectively) were lower than those showing differential expression between hybrids and Coffea canephora (15.2% and 19.6 % for the comparison E C versus E and C E versus E, respectively, P < 0.001, Fisher’s exact tests). In each of the four comparisons of hybrids with their parents, the percentage of upregulated genes was higher in the hybrids than in the parents (3.8% and 15.4% of upregulated genes in C E compared with 1.6% and 4.2% in C. eugenioides and C. canephora, respectively, 2.5% and 11.8% of upregulated genes in E C compared with 1.6% and 3.4% in C. eugenioides and C. canephora, respectively, P < 0.001, Fisher’s exact tests).
Combes et al.
GBE
trans-regulatory divergences (Cowles et al. 2002; Wittkopp et al. 2004). In our experimental conditions, half the genes showed no expression divergence between parental species and no allelic expression changes in hybrids. In the remaining genes, divergence in allelic expression was classified in three categories with no clear majority among the three: on aver-age, 15.5% in the cis-regulatory category alone, 18.5% in the regulatory category alone, and 17.5% in the cis- + trans-regulatory categories (fig. 4B), in the two groups of hybrids. The absolute magnitude of parental gene expression diver-gence in each regulatory diverdiver-gence category was evaluated. Trans effects appeared to play a larger role than cis effects in parental gene expression divergence in both groups of hybrids (fig. 4C). The data indicate that divergent evolution between diploid species may have impacted the regulatory sequences of genes (cis-regulatory factors) and abundance, as well as the activity of upstream regulators throughout the genome (trans-regulatory factors). In addition, the data also provided
evidence for the compatibility of diverged parental regulatory networks and of intertwined regulation of alleles. Indeed for all genes showing trans-regulatory expression divergence (combined with cis-regulatory or not), ASE was controlled by trans-regulatory factors from both parental genomes that are able to cause up- or downregulation of one or both alleles. Intertwined regulation of parental species has been already reported in gene expression in C. arabica (Combes et al. 2013). In several studies considering intra- or interspecific hy-brids of variable organisms (flies, yeast, plants), cis-regulatory divergences predominated and explained more of the expres-sion differences between species than within species (Emerson and Li 2010;Shi et al. 2012). With the exception of the study ofMcManus et al. (2010),trans-regulatory divergences con-tributed preferentially to the expression differences within spe-cies (Emerson et al. 2010;Bell et al. 2013;Xu et al. 2014). Others observations, such as the relative proportion of genes with evidence of cis- and/or trans-regulatory divergences,
FIG. 3.—Inheritance of gene expression level in hybrids between Coffea canephora and Coffea eugenioides. (A) Hypothetical patterns of gene expression in C. canephora (C), C. eugenioides (E), and F1hybrids; C E or E C (hyb) illustrating additivity, transgressivity, and dominance are shown. (B) For the
21,052 genes in C E and E C hybrids examined, the scatterplots compare the level of gene expression of F1 hybrids and those of the parental species (C. canephora on the x axis and C. eugenioides on the y axis). (C) The bar graphs show the frequency of genes in each expression inheritance category for C E and E C hybrids.
Regulatory Divergence between Parental Alleles
GBE
were more variable among studies. ForCoolon et al. (2014)
who studied expression differences within and between spe-cies of Drosophila, the proportion of genes with cis-regulatory divergences increases with divergence time but also depends on the mode of regulatory evolution of the species. In the study of Drosophila melanogaster and Drosophila sechellia hy-brids,McManus et al. (2010)interpreted the higher propor-tion of trans-regulatory divergences than expected for an interspecific hybrid as an effect of particular demographic and ecological history of D. sechellia (McManus et al. 2010). In the case of Coffea species, gene expression differences result notably from cis-regulatory divergences but also from
a larger part than expected of trans-regulatory divergences associated or not with cis-regulatory divergences. This finding could be related to the recent speciation and the low genetic divergence between parental species. In addition, the peren-nial habit of Coffea species and the long generation time (i.e., 10–20 years) of coffee trees in natural conditions associated with a slow molecular evolution rate could be considered (Cenci et al. 2013).
The GO term enrichments of genes belonging to cis- or trans-regulatory divergences categories have revealed that GO terms related to “cellular metabolic process” were over represented among genes displaying cis-regulatory changes,
FIG. 4.—cis- and trans-regulatory divergence between parental species for 11,438 genes in C E hybrids and 11,270 genes in E C hybrids. (A) The
plots summarize the relative allele-specific expression levels in parents and hybrids (C/E on the x axis; Chyb/Ehybon the y axis; C, E and Chyb, Ehybcorresponding
to parental and allelic read counts, respectively). Each point represents a single gene on a logarithmic scale, and is color coded according to the regulatory divergence categories inferred from a hierarchical series of statistical tests (see Material and Methods). (B) The bar graphs depict the frequency of genes in each regulatory divergence category for C E and E C hybrids. The exact number of genes in each category is indicated at the extremity of each bar. The numbers in bold on the bar indicate the percentage of genes relative to the total number of genes analyzed. (C) The box plots summarize the absolute magnitude (fold-change) of parental expression divergence resulting from cis, trans, cis and trans (compensating or enhancing) interactions. In C E and E C hybrids, each category of regulatory divergence showed increased levels of expression divergence (Wilcoxon’s rank-sum test, P < 0.001). The median significant trans-regulatory difference between species (1.15-fold for C E and 1.22-fold for E C hybrids) was larger than the median cis-regulatory difference between species (0.88-fold for C E and 0.95-fold for E C hybrids, Wilcoxon’s rank-sum test, P < 0.001).
Combes et al.
GBE
while those associated with “regulation of cellular metabolic process,” development, and response to stimulus were over represented among genes displaying trans-regulatory changes (supplementary table S4,Supplementary Materialonline). This result is consistent with the observations ofTirosh et al. (2009)
in yeast. Indeed, these authors showed that divergent expres-sion of genes responding to the environment may be prefer-entially caused by trans-regulatory changes. They also established that trans effects were attributable to differential interpretation of sensory signals and not to mutations in direct transcription regulators (Tirosh et al. 2009). Therefore, the observed trans-regulatory divergences between Coffea spe-cies could contribute to their different abilities to growth in contrasted environments. A study of gene expression of these same hybrids in variable controlled environments could be suggested to explore the phenomenon of Coffea species adaptation.
In this study, the analysis of the fraction of enhancing (cis and trans effects in the same direction) versus compensating interactions (cis and trans effects in opposite directions) in the category of cis- + trans-regulatory divergences makes possible to predict evolutionary models. Although stabilizing selection would be characterized by compensating cis and trans effects, diversifying selection would be revealed by enhancing cis and trans effects (Tirosh et al. 2009; Shi et al 2012). For both groups of hybrids, the fractions of genes with compensating cis- + trans-regulatory effects among genes with cis- + trans-regulatory divergences were dominant (prop.test, P < 0.001). In addition, the absolute magnitude of parental gene expression divergence in this category was quite similar to the absolute magnitude of parental gene expression diver-gence in cis- and trans-regulatory categories. These data suggest a prevalent role for stabilizing selection that maintains gene expression levels.
A Clear Relationship between Regulation of Allelic Expression and Expression Inheritance in Hybrids
In hybrids, trans-regulatory factors from both parental origins could bind cis-regulatory regions of alleles differently and have different effects on expression between alleles. Analysis of the expression variation of alleles from both genomes between the parental and the hybrid contexts reveals the roles of trans-regulatory factors, and hence should elucidate the evolution of the parental species (Shi et al. 2012). To further analyze regulation of alleles in the differentially expressed genes in the hybrids compared with the parental species, we assessed var-iation in ASE in the two groups of hybrids. For each gene, we estimated variations in ASE in either the C genome (C-ASE) with the RAE Chyb/C or the E genome (E-ASE) with the RAE
Ehyb/E between the parental and hybrid contexts (fig. 5and
supplementary fig. S2, Supplementary Material online, for C E and E C hybrids, respectively). Given that the genome-specific cis-regulatory factors are similar in the
parents and the hybrids, trans-regulatory effects on the varia-tion in ASE were revealed as well as their genome origin. The RAE of the C genome (Chyb/C) was plotted against the RAE of
the E genome (Ehyb/E) on a logarithmic scale. Log2positive
values of the RAE Chyb>C or Ehyb>E indicate upregulation
of the alleles by the trans-regulatory factors of the homolo-gous genome, while negative values of this ratio indicate downregulation of the alleles. The gene position relative to both axes reveals the single or combined genome origin of the trans-regulatory factors acting on that gene. Although alleles of both C and E genomes displayed both upregulated and downregulated allelic expression, the effects of trans-regulatory factors on both genomes appeared to be genome dependent. The Ehyballeles were up- and
downreg-ulated by C trans-regulatory factors, whereas Chyballeles were
mainly upregulated by E trans-regulatory factors. What is more, the E trans-regulatory factors had a greater impact on C-ASE than the C trans-regulator factors on E-ASE; the mag-nitude of variation in the expression of the Chyballele was
higher than that of the Ehyballele. These results suggest
asym-metric effects of the activation of alleles of one genome by trans-regulatory factors of the other.
To refine this observation, the effects of C and E trans-regulatory factors on the Chyb and Ehyb alleles in hybrids
were analyzed. For genes in the cis- + trans-regulatory catego-ries and trans-regulatory divergence categocatego-ries, Chyband Ehyb
alleles were classified according to the magnitude of the var-iation in allele expression between the hybrid and the parents (fig. 6). More Ehyballeles than Chyballeles displayed only slight
changes in ASE between the hybrids and their parental spe-cies, whereas more Chyb alleles than Ehyb alleles showed
marked ASE changes. These observations confirm that the E trans-regulatory factors have greater effects than C trans-reg-ulatory factors on ASE variation between parental species and hybrids, in agreement with the observed overall E expression level dominance. As in the study ofHe et al. (2012)where asymmetry of ASE in Arabidopsis hybrids has been explained by differences in the efficiency of epigenetic gene silencing (He et al. 2012), one can also imagine epigenetic causes of the asymmetric effects of the trans-regulatory factors of one genome on the activation of alleles of the other. Indeed, the more effective E trans-regulatory factors observed in both the C E and E C hybrids for genes showing trans-regulatory divergences could indicate divergent epigenetic states be-tween the two parental species.
To elucidate the regulatory mechanisms at the root of the gene expression inheritance, the up- and downregulation of the expression of alleles and the expression inheritance of genes were interpreted jointly (fig. 5 for C E and supple-mentary fig. S2,Supplementary Materialonline, for E C hy-brids). For the majority of genes classified in the E expression level dominance “up” category, the Chyballeles were
upreg-ulated by E trans-regulatory factors, whereas for genes in the E expression level dominance “down” category, the Chyballeles
Regulatory Divergence between Parental Alleles
GBE
FIG. 5.—Variation in specific allelic expression (ASE) for genes differentially expressed between parents and the hybrid in the C E hybrids. Each point
represents a single gene, its position displays the combined up- and downregulation of both alleles (Chyb/C on the x axis; Ehyb/E on the y axis, Chyb, Ehyband C,
E corresponding to allelic and parental read counts, respectively), and its color code, the inheritance category of the gene. In this way, the combined regulation of alleles at the root of expression inheritance was elucidated for each gene analyzed (chi-square test, P < 0.001).
FIG. 6.—Effects of trans-regulatory factors of the E and C genome on up- and downregulation of Chyband Ehyballeles in C E and E C hybrids. In line
with the method ofShi et al. (2012),the genes subjected to either only trans- or only cis + trans effects were categorized according to differences in the absolute values of allelic expression between C E and E C hybrids and each parental species (on the x axis, dark gray for Chyb/C and light gray for Ehyb/E).
Although more Ehyballeles showed a low level of expression difference between the parents and the hybrid, more Chyballeles displayed a high level of
expression difference (chi-squared test, P < 0.001).
Combes et al.
GBE
were downregulated. The same was observed for the C expression level dominance up and down categories, the Ehyb alleles being up- and downregulated by the C
trans-regulatory factors. For the additivity and transgressivity cate-gories, gene expression appeared to result from the combined effects of E and C trans-regulatory factors. Each category of expression inheritance of genes in hybrids depends on partic-ular combinations of up- and downregulation of both alleles. In agreement with two previous studies that investigated both homoeologous and total gene expression in allopolyploids (Yoo et al. 2013;Cox et al. 2014), the present data confirmed
that the biased expression level dominance toward one parent was mainly caused by up- or downregulation of the allele of the other parent. In addition, in this study, the observed biased expression level dominance was attributed to asymmetric ef-fects of trans-regulators factors of parental genomes. Similarly, between Arabidopsis thaliana and Arabidopsis are-nosa, Shi et al. (2012) observed higher effects of trans-regulatoty factors of A. arenosa on allelic expression in hybrids. Moreover, an overall expression dominance of A. arenosa genes over A. thaliana genes was previously reported in syn-thetic allopolyploids between A. arenosa and A. thaliana
FIG. 7.—For the C E hybrids, all the data collected on genes belonging to each category of expression inheritance are combined. The distributions of
genes according to RAE and to cis- and trans-regulatory divergences are shown in the form of histograms and pie charts.
Regulatory Divergence between Parental Alleles
GBE
(Wang et al. 2006). For other studies in various biological organisms (Bell et al. 2013; Yoo et al. 2013; Cox et al. 2014), the phenomenon of expression level dominance would be also attributed to the unbalanced effects of parental trans-regulatory factors. Accordingly, we hypothesize that in hybrids, gene expression inheritance could depend on the effects of trans-regulatory factors of one genome on the other genome, and hence the regulatory divergence between parental species. Thus, the divergent evolutionary histories of parental species could partly explain the different patterns of gene expression inheritance in hybrids reported in previous studies (McManus et al. 2010;Shi et al. 2012;Bell et al. 2013). Particular RAE Distribution for Each Category of
Expression Inheritance
Next, we refined the description of regulatory mechanisms of differentially expressed genes in interspecific hybrids relative to their parental species in the two groups of hybrids. For each category of expression inheritance, the distribution of genes according to RAE and regulatory divergence were compared with the distribution of genes in the conserved genes cate-gory, and significant differences were revealed (fig. 7for C E andsupplementary fig. S3,Supplementary Materialonline, for E C hybrids). Whatever the category of expression inheri-tance, the distributions of RAE were always unbalanced and biased toward the parental allele corresponding to the parent more expressed for the considered category. In the transgres-sivity up category, in which the two parental species were not differentially expressed, the RAE was more balanced than in other categories. With regard to the distributions of regulatory divergence mechanisms, the proportion of genes classified in each category varied between the categories of expression inheritance. The data showed that both cis- and trans-regula-tory divergences were associated with variable allelic expres-sion and consequently involved in variations of gene expression in hybrids. These findings reveal the intertwined allelic expression regulation and the complex relationship be-tween the cis- and trans-regulatory factors that modulate gene expression. In addition, trans-regulatory divergences be-tween parental species can have different origins: Changes in chromatin structures but also in transcription factors, changes in effectors, and RNA interference (RNAi) (Groszmann et al. 2011;Chodavarapu et al. 2012;He et al. 2012;Shen et al. 2012;Donoghue et al. 2014;Greaves et al. 2014).
Conclusions
Our genome-wide analysis enabled us to characterize both the genetic divergence between parental species and new gene expression patterns in hybrids. Gene expression changes between C. canephora and C. eugenioides resulted from both cis- and trans-divergences. The importance of trans-regulatory divergences observed between these two species could be related to their low genetic divergence and perennial habit.
In Coffea hybrids, the pattern of gene expression inheritance was characterized by a genome-wide expression level domi-nance biased toward the C. eugenioides parent. Moreover, gene expression was shown to result from the expression of both alleles and was affected by intertwined parental trans-regulatory factors indicating compatibility of parental regula-tory networks. Hence, gene expression inheritance appeared to be determined by a complex mix of cis-regulatory factors and asymmetric effects of divergent trans-regulatory parental factors on up- and downregulation of alleles. In particular, the phenomenon of biased expression level dominance was at-tributed to unbalanced effects of parental trans-regulatory factors. More generally in hybrids, the extent and relative rates of cis- and trans-divergence among the parental species would have a significant effect on the gene expression pat-tern. In addition, the more similar the parental genomes, the more likely it is that the transcription factors of one genome might be compatible with transcription factor binding sites on the other genome. It appears that in hybrids, the characteris-tics of the transcriptional response to genome merger depend on genetic divergences between the parental species and their evolutionary history.
Supplementary Material
Supplementary figures S1–S3,tables S1–S4, and Methods are available at Genome Biology and Evolution online (http:// www.gbe.oxfordjournals.org/).
Acknowledgment
This work was partially supported by a grant from the National Centre for Coffee Research and the Agriculture Minister of Colombia (contract FNC 217/2011).
Literature Cited
Ames RM, Lovell SC. 2011. Diversification at transcription factor binding sites within a species and the implications for environmental adapta-tion. Mol Biol Evol. 28:3331–3344.
Anders S, Huber W. 2010. Differential expression analysis for sequence count data. Genome Biol. 11:R106.
Anthony F, Diniz LEC, Combes M-C, Lashermes P. 2010. Adaptive radia-tion in Coffea subgenus Coffea L. (Rubiaceae) in Africa and Madagascar. Plant Syst Evol. 285:51–64.
Barrier M, Robichaux RH, Purugganan MD. 2001. Accelerated regulatory gene evolution in an adaptive radiation. Proc Natl Acad Sci U S A. 98: 10208–10213.
Bell GD, Kane NC, Rieseberg LH, Adams KL. 2013. RNA-seq analysis of allele-specific expression, hybrid effects, and regulatory divergence in hybrids compared with their parents from natural populations. Genome Biol Evol. 5:1309–1323.
Buggs RJ, et al. 2014. The legacy of diploid progenitors in allopolyploid gene expression patterns. Philos Trans R Soc Lond B Biol Sci. 369. pii:20130354.
Cenci A, Combes MC, Lashermes P. 2012. Genome evolution in diploid and tetraploid Coffea species as revealed by comparative analysis of orthologous genome segments. Plant Mol Biol. 78:135–145.
Combes et al.
GBE
Cenci A, Combes M-C, Lashermes P. 2013. Differences in evolution rates among eudicotyledon species observed by analysis of protein diver-gence. J Hered. 104:459–464.
Chague´ V, et al. 2010. Genome-wide gene expression changes in genet-ically stable synthetic and natural wheat allohexaploids. New Phytol. 187:1181–1194.
Chapman MA, Hiscock SJ, Filatov DA. 2013. Genomic divergence during speciation driven by adaptation to altitude. Mol Biol Evol. 30: 2553–2567.
Chodavarapu RK, et al. 2012. Transcriptome and methylome interactions in rice hybrids. Proc Natl Acad Sci U S A. 109:12040–12045. Combes M-C, Dereeper A, Severac D, Bertrand B, Lashermes P. 2013.
Contribution of subgenomes to the transcriptome and their inter-twined regulation in the allopolyploid Coffea arabica grown at con-trasted temperatures. New Phytol. 200:251–260.
Conesa A, Gotz S. 2008. Blast2GO: a comprehensive suite for functional analysis in plant genomics. Int J Plant Genomics. 2008:619832. Conesa A, et al. 2005. Blast2GO: a universal tool for annotation,
visuali-zation and analysis in functional genomics research. Bioinformatics 21: 3674–3676.
Coolon JD, McManus CJ, Stevenson KR, Graveley BR, Wittkopp PJ. 2014. Tempo and mode of regulatory evolution in Drosophila. Genome Res. 24:797–808.
Couturon E, Lashermes P, Charrier A. 1998. First intergeneric hybrids (Psilanthus ebracteolatus Hiern Coffea arabica L.) in coffee trees. Can J Bot. 76:542–546.
Cowles CR, Hirschhorn JN, Altshuler D, Lander ES. 2002. Detection of regulatory variation in mouse genes. Nat Genet. 32:432–437. Cox MP, et al. 2014. An interspecific fungal hybrid reveals cross-kingdom
rules for allopolyploid gene expression patterns. PLoS Genet. 10: e1004180.
Cros J, et al. 1998. Phylogenetic analysis of chloroplast DNA variation in Coffea L. Mol Phylogenet Evol. 9:109–117.
Denoeud F, et al. 2014. The coffee genome provides insight into the convergent evolution of caffeine biosynthesis. Science 345: 1181–1184.
Donoghue MTA, et al. 2014. C(m)CGG methylation-independent parent-of-origin effects on genome-wide transcript levels in isogenic recipro-cal F1 triploid plants. DNA Res. 21:141–151.
Emerson JJ, Li WH. 2010. The genetic basis of evolutionary change in gene expression levels. Philos Trans R Soc Lond B Biol Sci. 365:2581–2590. Emerson JJ, et al. 2010. Natural selection on cis and trans regulation in
yeasts. Genome Res. 20:826–836.
Greaves IK, Groszmann M, Wang A, Peacock WJ, Dennis ES. 2014. Inheritance of trans chromosomal methylation patterns from Arabidopsis F1 hybrids. Proc Natl Acad Sci U S A. 111:2017–2022. Groszmann M, et al. 2011. Changes in 24-nt siRNA levels in Arabidopsis
hybrids suggest an epigenetic contribution to hybrid vigor. Proc Natl Acad Sci U S A. 108:2617–2622.
Grover CE, et al. 2012. Homoeolog expression bias and expression level dominance in allopolyploids. New Phytol. 196:966–971.
He F, et al. 2012. Genome-wide analysis of cis-regulatory divergence be-tween species in the Arabidopsis genus. Mol Biol Evol. 29:3385–3395.
Hegarty MJ, Hiscock SJ. 2009. The complex nature of allopolyploid plant genomes. Heredity 103:100–101.
Kapralov MV, Votintseva AA, Filatov DA. 2013. Molecular adaptation during a rapid adaptive radiation. Mol Biol Evol. 30:1051–1059. Landry CR, Hartl DL, Ranz JM. 2007. Genome clashes in hybrids: insights
from gene expression. Heredity 99:483–493.
Lashermes P, et al. 1999. Molecular characterisation and origin of the Coffea arabica L. genome. Mol Gen Genet. 261:259–266.
Li H, Durbin R. 2010. Fast and accurate long-read alignment with Burrows-Wheeler transform. Bioinformatics 26:589–595.
Louarn J. 1976. Hybrides interspe´cifiques entre Coffea canephora pierre et C. eugenioides Moore. Cafe´ Cacao The´. 20:433–452.
Marioni JC, Mason CE, Mane SM, Stephens M, Gilad Y. 2008. RNA-seq: an assessment of technical reproducibility and comparison with gene expression arrays. Genome Res. 18:1509–1517.
McManus CJ, et al. 2010. Regulatory divergence in Drosophila revealed by mRNA-seq. Genome Res. 20:816–825.
Muir G, Osborne OG, Sarasa J, Hiscock SJ, Filatov DA. 2013. Recent eco-logical selection on regulatory divergence is shaping clinal variation in senecio on Mount Etna. Evol Int J Org Evol. 67:3032–3042. Noirot M, et al. 2003. Genome size variations in diploid African Coffea
species. Ann Bot. 92:709–714.
Qi B, et al. 2012. Global transgenerational gene expression dynamics in two newly synthesized allohexaploid wheat (Triticum aestivum) lines. BMC Biol. 10:3.
Shen H, et al. 2012. Genome-wide analysis of DNA methylation and gene expression changes in two Arabidopsis ecotypes and their reciprocal hybrids. Plant Cell 24:875–892.
Shi X, et al. 2012. Cis- and trans-regulatory divergence between progenitor species determines gene-expression novelty in Arabidopsis allopoly-ploids. Nat Commun. 3:950.
Tirosh I, Reikhav S, Levy AA, Barkai N. 2009. A yeast hybrid provides insight into the evolution of gene expression regulation. Science 324: 659–662.
Wang L, Wang S, Li W. 2012. RSeQC: quality control of RNA-seq exper-iments. Bioinformatics 28:2184–2185.
Wang J, et al. 2006. Genomewide nonadditive gene regulation in Arabidopsis allotetraploids. Genetics 172:507–517.
Whitehead A, Crawford DL. 2006. Variation within and among species in gene expression: raw material for evolution. Mol Ecol. 15:1197–1211. Wittkopp PJ, Haerum BK, Clark AG. 2004. Evolutionary changes in cis and
trans gene regulation. Nature 430:85–88.
Wolf JBW, Lindell J, Backstro¨m N. 2010. Speciation genetics: current status and evolving approaches. Philos Trans R Soc B Biol Sci. 365: 1717–1733.
Xu C, et al. 2014. Genome-wide disruption of gene expression in allopoly-ploids but not hybrids of rice subspecies. Mol Biol Evol. 31:1066–1076. Yoo M-J, Szadkowski E, Wendel JF. 2013. Homoeolog expression bias and expression level dominance in allopolyploid cotton. Heredity 110: 171–180.
Associate editor: Patricia Wittkopp
Regulatory Divergence between Parental Alleles
